Re: [AMBER] MMPBSA: Default inp warnings

From: Jason Swails <>
Date: Wed, 17 Sep 2014 13:57:24 -0400

On Wed, 2014-09-17 at 14:39 -0300, George Tzotzos wrote:
> I’m using to calculate binding energies. The script I’m
> using is:
> Input file for running PB and GB
> &general
> interval=2, endframe=10000, verbose=1,
> # entropy=1,
> /
> &gb
> igb=1, saltcon=0.100
> /
> &pb
> inp=1, istrng=0.100,
> /
> Using inp=1 throws out warning messages indicating that inp=1 is the
> old default

I think the warnings are actually stating that the defaults for _other_
parameters in the &pb namelist are applicable to the inp=2 model, and
that they are being changed to the defaults for the inp=1 model that you
chose. So this is fine.
> Running the same calculation with inp=2 and comparing the results the
> discrepancies for the Poisson Bolzmann binding energies are pretty
> huge

This has been reported a number of times on this list. If inp=2 is
giving you physically unrealistic results, inp=1 may be the better model
to use for your system -- this has been discussed quite a bit on the
Amber list in the past.
> GENERALIZED BORN (inp=1) -22.3955 kcal/mol -22.7632 kcal/mol (Complex
> - Receptor - Ligand)
> POISSON BOLTZMANN (inp=1) -18.7747 kcal/mol 0.8206 kcal/mol (Complex
> - Receptor - Ligand)

inp=1 is PB only. The GB nonpolar model is basically only like inp=1 (a
simple SASA-based method that is purely unfavorable). I'm not quite
sure what the numbers you posted are supposed to mean. I'm guessing the
first is inp=1 and the second inp=2? If so, I'm surprised the GB
energies changed at all, and the PB energies certainly indicate a
difference in the nonpolar binding free energies. Neither is probably a
great choice for absolute binding affinities, but hopefully relative
binding affinities show far less dependence on the nonpolar model you


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Wed Sep 17 2014 - 11:00:03 PDT
Custom Search